A proposed Framework for Aspect-Based Sentiment Analysis for Arabic Content using Semantic-Based Approach

Abstract

This research proposed a new framework for analyzing the Arabic opinions and measuring the sentiment analysis for Arabic contents using semantic approach and text analysis. This thesis focusing on the semantic model ontology engineering , text tagging and part of speech tagging. To prove the potential long-term advantages of this approach. The framework was successfully designed and developed and the process had been described as five components , first one is Arabic Part Of Speech “POS” tagger was used to assign the correct tag for every word in the opinion , second component is Arabic Ontology Classifier to query RDF Ontology using ontology engineering ; SPARQL language to extract the main concepts ,the third component is Arabic Sentiment Lexicon which act like Arabic dictionary and the final component is the Counter and Report component which do the calculation and final results and reports display, All previous components worked together as one solid framework called SUSTASA. SUSTASA framework had been designed and developed using advance integration technologies such as Stanford POS tagger and Stanford Ontology protégé. Jean framework from Apache had been used to design the Ontology. The framework tested against Arabic comments using the Arabic movie ontology, and the results from framework was successfully able to detect and classify the pilot comments written in Arabic language and measure them against the five star rating system . The results also agreed well with the calculated stars which entered by the opinion holders. The framework had been uploaded into GitHup and it’s available through internet as open source framework for scientists.